Simultaneous detection of omicron and other SARS-CoV-2 variants by multiplex PCR MassARRAY technology

The rapid emergence of SARS-CoV-2 variants with high severity and transmutability adds further urgency for rapid and multiplex molecular testing to identify the variants. A nucleotide matrix-assisted laser-desorption-ionization time-of-flight mass spectrophotometry (MALDI-TOF MS)-based assay was developed (called point mutation array, PMA) to identify four major SARS-CoV-2 variants of concern (VOCs) including Alpha, Beta, Delta, and Omicron (namely PMA-ABDO) and differentiate Omicron subvariant (namely PMA-Omicron). PMA-ABDO and PMA-Omicron consist of 24 and 28 mutation sites of the spike gene. Both PMA panels specifically identified VOCs with as low as 10 viral copies/µl. The panel has shown a 100% concordant with the Next Generation Sequencing (NGS) results testing on 256 clinical specimens with real-time PCR cycle threshold (Ct) values less than 26. It showed a higher sensitivity over NGS; 25/28 samples were positive by PMA but not NGS in the clinical samples with PCR Ct higher than 26. Due to the mass of nucleotide used to differentiate between wild-type and mutation strains, the co-infection or recombination of multiple variants can be determined by the PMA method. This method is flexible in adding a new primer set to identify a new emerging mutation site among the current circulating VOCs and the turnaround time is less than 8 h. However, the spike gene sequencing or NGS retains the advantage of detecting newly emerged variants.


Results
Performance of PMA. The mutation sites at the spike gene were chosen as a detection target to characterize the SARS-CoV-2 VOCs in this study. Table 1 shows a list of the mutation site and prevalence across lineages used in this study. It should be noted that L18F was included to differentiate the Gamma variant as its prevalence was 97.9%. The SARS-CoV-2 nucleoprotein (N) gene target was included in the PMA panels as the internal control to validate the assay sensitivity and performance. The PMA-ABDO panel consists of 24 specific mutation sites of the spike gene for detecting 4 VOCs (Alpha, Beta, Delta, and Omicron) simultaneously in one reaction ( Fig. 2A). The total gene target was 25 (Table 1), consisting of 25 extension primers with 22 forward and 21 reverse primers (available upon request). The second panel, the PMA Omicron panel, consists of 28 spike gene targets in which 20 targets are uniquely specific to Omicron. It can differentiate BA.2 from BA.1 and BA.1.1 variants in one reaction (Fig. 2B). The total gene target for PMA-Omicron is 29 ( Table 2), consisting of 29 extension primers with 27 forward and 27 reverse primers (available upon request).
The PMA cutoff was calculated from the mean % call rate of five viral isolates at 5 dilutions (10-10 6 copies/ µl) using GraphPad Prism 9.0 version (Supplement Table 1). It was determined as 43.59%, with 100% sensitivity (95% confidence interval (CI) [86.68%, 100%]) and 100% specificity (95% CI [83.89%, 100.0%]). In this study, the PMA validation criteria consist of (1) positive on the N gene and (2) the total % call rate higher than the cutoff, 43.59%. www.nature.com/scientificreports/ Specificity of the PMA method. Five viral isolates (1000 copies/µl) of SARS-CoV-2 ancestral, Alpha, Beta, Delta, and Omicron strains were used to evaluate the specificity of PMA-ABDO and PMA-Omicron assays. All mutation sites were identified correctly in all viral isolates for PMA-ABDO and PMA-Omicron panels (Supplement Tables 2 and 3). Twenty respiratory samples collected from non-COVID-19 patients who tested negative for SARS-COV-2 RT-PCR and other common respiratory viruses were used to evaluate non-specific signal of the PMA-ABDO assay. No call rate was detected on all 24 mutation sites and N gene in all samples; only unextended primer (UEP) signals were found (Supplement Table 4).

Sensitivity and specificity comparisons between PMA and NGS methods. WGS is considered
the gold standard for SARS-CoV-2 variant identification. A total of 256 SARS-CoV-2 positive samples (PCR cycle threshold (Ct) values of N gene less than 26) were analyzed using PMA-ABDO or PMA-Omicron, and the results were compared with those obtained by the NGS method (Group 1, ND not detected, Bold mutation site targets were used in both PMA panels. *These two mutation site used the same set of primer for amplification and extension. **These two mutation site used the same set of primer for amplification and extension. $ L18F is the hallmark mutation site for SARS-CoV-2 Gamma variant (97.9% prevalence). # Data on 1 May 2022, outbreak.info.

Mutation site Amino acid
Mutation sites nucleotides  Table 3). Three samples that failed to be detected by both methods were those with PCR Ct higher than 35. They contained at two unique mutation points (A570D and T716I), permitting them to be characterized as suspected Alpha variant (Group 2, Table 3). Twenty-three samples were identified as non-VOCs lineage by NGS, including one of A.6, B.  Table 3). N gene and wildtype nucleotides of 24 mutation sites of the spike gene were detected in all samples by PMA-ABDO panel.
Detection of heterozygous spectra. A mixed of Delta and Omicron mock clinical sample was prepared and used to test the ability of MassARRAY technology to distinguish multiple-strains infections. PMA-Omicron was used for the evaluation, and it was found that the D614G site appeared as a single peak since both Delta and Omicron variants have the same nucleotide mutation at this site. The heterozygous spectra were detected and interpreted as co-infection between Delta and Omicron at Q493R (A and G, respectively) and N856K (C and A, respectively) mutation sites (Fig. 3).

Discussion
Global genomic surveillance for real-time identification of the SARS-CoV-2 variant is a critical public health concern 11 . Whole-genome or complete spike gene sequencing is the gold standard for identifying the strain of SARS-CoV-2. Rapid detection methods with simultaneous detection of variants using real-time multiplex PCR are commercially available and widely used to screen SARS-CoV-2 variants before sequencing confirmation www.nature.com/scientificreports/ by NGS [12][13][14] . In a recent study, a similar technology (Multiplex PCR-Mass spectrometry Minisequencing) was developed for the differentiation of three SARS-CoV-2 VOCs, including Alpha, Beta, and Delta, using 9 mutation sites of the spike gene 10 . The detection limit of all 9 sites was 1.5 × 10 3 copies. In the other study, nucleic acid mass spectrometry has been previously developed for the simultaneous detection of seven human coronaviruses using 20 viral targets for differentiation; it showed the same sensitivity as that of Real-time PCR assay 15 . However, this method is difficult to identify the new human CoV or SARS-CoV-2 variant because the gene fragments with conserved intraspecies specificity were selected for amplification and detection. In our study, the MassARRAY technology to detect SARS-CoV-2 VOCs was expanded from the previous study 10 to include the Omicron variant and increase the number of targets to increase the specificity of the assay.

SARS-CoV-2 variant NGS examination (n)
PMA panel assay www.nature.com/scientificreports/ Two panels of PMA assay based on the MassARRAY technology were developed to simultaneously detect 24 mutation sites of the SARS-CoV-2 spike gene to identify four VOCs, including Alpha, Beta, Delta, and Omicron, called PMA-ABDO ( Fig. 2A). At the same time, a second assay called PMA-Omicron was developed to distinguish Omicron BA.2 from Omicron B.1.1.529 and BA.1 (Fig. 2B). The virus isolates were used to determine the limit of detection and specificity. The overall detection limit of PMA-ABDO against ancestral strain was 10 copies/µl except at P681R/H, where the detection limit was 1000 copies/µl. The detection limit of MassARRAY technology is affected by both multiplex PCR amplification and multiplex primer extension reactions 10 . Another study has shown that the binding efficiency of the extension primer was reduced when the mutation sites were located in the same PCR fragment, for example, an increase in a detection limit of P681R/H to 1560 copies compared to 400 copies of D614G 10 . A similar result was found in our study, the detection limit of P681R/H tested with five virus isolates; ancestral, Alpha, Beta, Delta, and Omicron strains, was 1000 copies/µl, while the detection limit of D614G and A701V were 10 copies/µl. This indicates that the binding efficiency of the P681R/H primer is likely to be lower than other mutation sites in the panel. The P681R/H mutation site was found in three SARS-CoV-2 variants, Alpha, Delta, and Omicron; its low sensitivity has a more negligible effect on the overall test performance. Two hundred fifty-six clinical samples of SARS-CoV-2 were used to compare the detection specificity between PMA and NGS. The detection limit of the NGS assay in our study was PCR Ct less than 26. The detection specificity of PMA-ABDO and PMA-Omicron of known variants and non-variant SARS-CoV-2 was 100% comparable to NGS results (Group 1, Table 3). Additionally, another study on multiplex PCR mass spectrometry has reported 100% accuracy in all samples with PCR Ct less than 27 10 . When testing 102 unknown SARS-CoV-2 variant clinical samples (PCR Ct 26-40) with PMA-ABDO and PMA-Omicron assays, 99 (97.05%) were tested positive (Group 2 and 3, Table 3). 25 of 28 samples (89.28%) were PMA positive while all were negative by NGS, indicating a higher sensitivity of PMA over NGS in the low viral copy samples. However, three samples with PCR Ct > 35 were tested positive for the SARS-CoV-2 N gene by PMA, but the variant types could not be identified due to the % call rate lower than the PMA cutoff (43.59%).
The non-specific binding of extension primer was not detected among Alpha, Beta, Delta, and Omicron variants and ancestral strains in all spike mutation sites included in PMA-ABDO (24 sites) and PMA-Omicron (28 sites) when tested with known virus isolates at 1000 copies/µl (Supplement Table 2 Table 4), were correctly identified by PMA-ABDO as non-COVID-19 according to our justification criteria. However, the non-specific peak (1-2 peaks per sample) was detected in the patient samples infected with other respiratory viruses (specimens were collected before 2019), but it was not found in the non-COVID-19 patients collected in 2021. This reveals that the non-specific detection was unlikely from host DNA binding. There are two possibilities for the cross-reactivity; firstly, the high number of primers in 25-plex PCR of PMA-ABDO may cause non-specific binding with non-targeted genes in other viruses. Secondly, other respiratory viruses, such as HCoV-229E, may contain parts of the genome similar to the targeted mutation sites where the primer can bind non-specifically. A recent study has shown that the Omicron variant harbors a unique insertion mutation of HCoV-229E 16 . The insertion mutation may be acquired through template switching involving the genome of HCoV-229E and SARS-CoV-2 in coinfected patients. Further study with more sample numbers is needed to confirm this observation.

and 3). Moreover, forty nucleic acids from clinical samples, including twenty of various non-SARS-CoV-2 respiratory viruses and 20 non-COVID-19 patients (Supplement
The limitation of variant screening tests, either by real-time PCR or multiplex PCR-Mass Spectrometry-based methods, is the ability to detect a new variant 10,12,14 . Due to circulating of multiple SARS-CoV-2 variants and rapid mutation, the multiple mutation sites should be included in the same assay or workflow to avoid misidentification. In the PMA-ABDO, the multiple lineage-defining markers were included for the identification of Alpha (A570D, T716I, S982A, D1118H, DEL144/144), Beta (D80A, D215G, K417N, A701V, and DEL241/243), Delta (D950N, DEL157/158, L452R, P681R), and Omicron (T95I, E484A) ( Fig. 2A). Moreover, target detection covering current VOCs in our PMA assay could detect the unusual recombinant variants such as Delta-Omicron or the XJ lineage (BA.1 and BA.2) or co-infection 17,18 . The mocked sample, a mixture of Delta and Omicron variants, was tested with PMA-Omicron assay, and the heterozygous spectra (Fig. 3) demonstrated the advantage of using the PMA system for distinguishing co-infections.
The PMA-Omicron used in this study contained the lineage-defining markers (T19I, DEL69/70, A67V, and Q493R) that can differentiate Omicron BA.3 and BA.4/BA.5. However, additional markers on ORF1a (DEL141/1433) or M gene (D3N) is needed to distinguish between BA.4 and BA.5 2 . In conclusion, the PMA method is a simple screening assay for SARS-CoV-2 variants with high sensitivity and specificity and has a highthroughput format (a maximum of 384 samples can be detected within 8 h). The PMA assay can detect mixed nucleotides at a single site (heterozygous) and show a promising tool for detecting SARS-CoV-2 recombination or co-infection associated with recognized VOC. The system is flexible, allowing primer sets to be adjustable with the maximum multiplex MPE up to 40 targets, which can help screen SARS-CoV-2 variants at a low cost.

Clinical specimens and virus isolates.
A total of 358 SARS-CoV-2 positive (Table 3) and 40 SARS-CoV-2 negative (Supplement Table 4) nasopharyngeal and throat (NT) swabs in viral transport media were used in this study. Detection of SARS-CoV-2 was determined by Allplex™ SARS-CoV-2 Real-time PCR Assay (Seegene Inc., Seoul, South Korea) of specific conserved fragments within RdRp, N, and E viral genes. The realtime PCR was conducted on Bio-Rad CSF96™ system according to the manufacturer's recommendation. Briefly, 17 μl of the One-step RT-PCR Mastermix was added into PCR tubes, and 8 μl of each sample's nucleic acids, 2019-nCoV positive control, and negative control (RNase-free Water) was added separately into the tubes. The amplification targets of RdRp, N, and E viral genes were detected through Cal Red 610, Quasar 670, and FAM www.nature.com/scientificreports/ channels, respectively. Whereas the internal control was detected with HEX channel. The cutoff PCR Ct was ≤ 40 and limit of detection was 100 RNA Copies/reaction. The PCR ct of N gene (less than 26) was used to determine sensitivities and the comparison with NGS. This study was approved by the Institutional Review Board of the Faculty of Medicine, Chulalongkorn University (Approval IRB No Primer design. Each PCR primer set consisted of three constituent primers, including two primers of firstround PCR (forward and reverse primer) and an extension primer for the detecting step (Fig. 1). The mutation sites were selected from variants reported on the mutation database outbreak.info platform 2 and the primers were designed from the reference SARS-CoV-2 genome (NC_045512. Shrimp alkaline phosphatase (SAP) treatment. The unincorporated dNTPs in the multiplex PCR reaction were removed by SAP. In brief, 2 µl of SAP enzyme mixture (iPLEX ® Pro, Agena Bioscience, CA, USA) was added to 6 µl PCR product and incubated at 37 °C for 40 min. The reactions were stopped by inactivation at 85 °C for 5 min.
Multiplex primer extension (MPE) reactions. Selected mutations on the spike gene (24 and 28 targets for PMA-ABDO and PMA-Omicron assays, respectively) and one N gene were analyzed in a single reaction. The MPE reaction was performed with terminator nucleotides (ddNTPs) and the designed extension primers were used to distinguish the mass difference of the mutant nucleotide from the wild-type (Fig. 1). The MPE reaction mixture (ddNTP mix, iPLEX ® Buffer Plus, iPLEX ® Enzyme, and mix mass extension primers) was added to the SAP treatment product, according to the manufacturer protocol. The extension reactions were carried out at 95 °C for 30 s and then 95 °C for 5 s, followed by 5 cycles at 52 °C for 5 s and 80 °C for 5 s, for a total of 40 cycles, and then 72 °C for 3 min. The final product was desalinated using a resin column.
MassARRAY data acquiring and analysis. Purified MPE products were transferred to a 96-well spectroCHIP of MassARRAY (Agena Bioscience, CA, USA) using MALDI-TOF mass spectrometry. Genotype Calling was performed in a real-time manner with MassARRAY RT software version 4.1 and analyzed with MassARRAY Typer software.
Determine cutoff, limit of detection and specificity. The known virus isolates, including ancestral, Alpha, Beta, Delta, and Omicron strains, were obtained from the Department of Virology, AFRIMS, Bangkok, and used to assess the cutoff, limit of detection (LOD), and specificity of the mutation panels. Ten-fold serial dilution of 10 to 10 6 copies/µl was used to determine the cutoff and LOD of the PMA assays, while the specificity was determined at a concentration of 1000 copies/µl. Each RNA sample was tested in triplicate. The % call rate was calculated from the number of positive peaks divided by the total number of expected peaks and multiplied by 100. www.nature.com/scientificreports/ The % call rate cutoff was calculated from the correlation between the amount of virus from 10 to 10 6 copies/ µl and the mean of % call rate from five viral isolates determined by MassARRAY, GraphPad Prism 9.0 was used for cutoff analysis (Supplement Table 1).

Data availability
The primer sequence data used in this study are available from Dr. Opass Putcharoen (opassid@gmail.com) but restrictions apply to the availability of these data, which were used under license (license application no. 2203001009) for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of Dr. Opass Putcharoen to any researcher wishing to use them for non-commercial purposes. Researchers who wish to obtain a copy of the data submit their request to opassid@ gmail.com.